BIOIMAGING 2016 Abstracts

Full Papers Short Papers
Full Papers
Paper Nr: 4
Title:

Automated Breast Mass Segmentation using Pulse-Coupled Neural Network and Distance Regularized Level Set Evolution: A Coarse-to-fine Approach
Authors:

Songlin Du, Yaping Yan and Yide Ma
Abstract: Motivation: Computer-aided diagnosis (CAD) is an important means for the clinical detection of breast cancer. Mass is a common manifestation of breast cancer. This work aims to develop an effective breast mass segmentation algorithm for CAD systems. Method: On one hand, pulse-coupled neural network (PCNN) and level set (LS) method have complementary advantages in image segmentation, we therefore combine PCNN and LS. On the other hand, traditional LS method formulates the evolution of the contour through the evolution of a level set function (LSF), and LSF typically develops irregularities during its evolution, which may cause numerical errors and eventually destroy the stability of the evolution. So we use an improved LS model, named distance regularized level set evolution (DRLSE), to achieve desirable segmentation performance. Specifically, we extract the region of interest (ROI) with PCNN and sets initial contour for DRLSE first. Then the finely segmentation is achieved by DRLSE. Results: Both qualitative and quantitative experiments on three large-scale mammography databases prove that the proposed method achieves high segmentation accuracy. Conclusion: The proposed algorithm is effective for automatic breast mass segmentation. Significance: First, the sketchy position of mass is fixed by PCNN, which guides the algorithm to define a flexibly initial contour for DRLSE. This strategy makes it easier for the contour to move from initial position towards the boundary between mass and normal tissue. Second, the use of DRLSE, which introduces an intrinsic capability of maintaining regularity of the LSF, ensures stable LS evolution and achieves accurate segmentation.
Download
Paper Nr: 9
Title:

SynapCountJ: A Tool for Analyzing Synaptic Densities in Neurons
Authors:

Gadea Mata, Jónathan Heras, Miguel Morales, Ana Romero and Julio Rubio
Abstract: The quantification of synapses is instrumental to measure the evolution of synaptic densities of neurons under the effect of some physiological conditions, neuronal diseases or even drug treatments. However, the manual quantification of synapses is a tedious, error-prone, time-consuming and subjective task; therefore, tools that might automate this process are desirable. In this paper, we present SynapCountJ, an ImageJ plugin, that can measure synaptic density of individual neurons obtained by immunofluorescence techniques, and also can be applied for batch processing of neurons that have been obtained in the same experiment or using the same setting. The procedure to quantify synapses implemented in SynapCountJ is based on the colocalization of three images of the same neuron (the neuron marked with two antibody markers and the structure of the neuron) and is inspired by methods coming from Computational Algebraic Topology. SynapCountJ provides a procedure to semi-automatically quantify the number of synapses of neuron cultures; as a result, the time required for such an analysis is greatly reduced.
Download
Paper Nr: 17
Title:

Tackling the Problem of Data Imbalancing for Melanoma Classification
Authors:

Mojdeh Rastgoo, Guillaume Lemaitre, Joan Massich, Olivier Morel, Franck Marzani, Rafael Garcia and Fabrice Meriaudeau
Abstract: Malignant melanoma is the most dangerous type of skin cancer, yet melanoma is the most treatable kind of cancer when diagnosed at an early stage. In this regard, Computer-Aided Diagnosis systems based on machine learning have been developed to discern melanoma lesions from benign and dysplastic nevi in dermoscopic images. Similar to a large range of real world applications encountered in machine learning, melanoma classification faces the challenge of imbalanced data, where the percentage of melanoma cases in comparison with benign and dysplastic cases is far less. This article analyzes the impact of data balancing strategies at the training step. Subsequently, Over-Sampling (OS) and Under-Sampling (US) are extensively compared in both feature and data space, revealing that NearMiss-2 (NM2) outperform other methods achieving Sensitivity (SE) and Specificity (SP) of 91.2% and 81.7%, respectively. More generally, the reported results highlight that methods based on US or combination of OS and US in feature space outperform the others.
Download
Paper Nr: 20
Title:

Reproducibility Analysis of 4DCT Derived Ventilation Distribution Data - An Application of a Ventilation Calculation Algorithm based on 4DCT
Authors:

Geoffrey G. Zhang, Kujtim Latifi, Vladimir Feygelman, Thomas J. Dilling and Eduardo G. Moros
Abstract: Deriving lung ventilation distribution from 4-dimensional CT (4DCT) using deformable image registration (DIR) is a recent technical development. In this study, we evaluated the serial reproducibility of ventilation data derived from two separate 4DCT data sets, collected at different time points. A total of 33 lung cancer patients were retrospectively analyzed. All patients had two stereotactic body radiotherapy treatment courses for lung cancer. Seven patients were excluded due to artifacts in the 4DCT data sets. The ventilation distributions in the lungs for each patient were calculated using the two sets of planning 4DCT data. The deformation matrices between the expiration and inspiration phases generated by DIR were used to produce ventilation distributions using the ΔV method. Ventilation in the lung regions that received less than 1 Gy was analyzed. For the 26 cases, the median Spearman correlation coefficient value was 0.31 (range 0.18 to 0.52, p value < 0.01 for all cases). The median Dice similarity coefficient value between the upper 30% ventilation regions of the two sets was 0.75 (range 0.71 to 0.81, Figure 1). We conclude that the two ventilation data sets in each case correlated and the reproducibility over time was reasonably good.
Download
Paper Nr: 27
Title:

Design of Customised Orthodontic Devices by Digital Imaging and CAD/FEM Modelling
Authors:

Sandro Barone, Alessandro Paoli, Armando Viviano Razionale and Roberto Savignano
Abstract: In recent years, the public demand of less invasive orthodontic treatments has led to the development of appliances that are smaller, lower profile and more transparent with respect to conventional brackets and wires. Among aesthetic appliances, removable thermoplastic aligners gained instant appeal to patients since able to perform comprehensive orthodontic treatments without sacrificing comfort issues. The aligner must deliver an appropriate force in order to move the tooth into the expected position. However, at present, the relationship between applied force and aligner properties (i.e., aligner’s thickness) is poorly understood. In this paper, a patient-specific framework has been developed to simulate orthodontic tooth movements by using aligners. In particular, a finite element model has been created in order to optimise the aligner’s thickness with regard to the magnitude of the force-moment system delivered to a mandibular central incisor during bucco-lingual tipping.
Download
Short Papers
Paper Nr: 13
Title:

Automatic Segmentation of Extensor Tendon of the MCP Joint in Ultrasound Images
Authors:

Malik Saad Sultan, Nelson Martins, Diana Veiga, Manuel Ferreira and Miguel Coimbra
Abstract: Rheumatoid arthritis (RA) is a chronic inflammatory disease that primarily affects the small joints of the hand. High frequency ultrasound imaging is used to measure the inflammatory activity in the joint capsule region of Metacarpophalangeal (MCP) joint. In our previous work, the problem of bones and joint capsule segmentation was addressed and in this work we aim to automatically identify the tendon using previously segmented structures. The extensor tendon is located above the metacarpal and phalange bone and the joint capsule. Tendon and bursal involvement are frequent and often clinically dominant in early RA. Ridge-like structures are enhanced and pre-processed to reduce speckle noise using a Log-Gabor filter. These regions are then simplified using medial axis transform and vertically connected lines are removed. Adjacent lines are connected using morphological operators and short lines are filtered by thresholding. Physiological information is used to create a distance map for all the lines using prior knowledge of the bone and capsule region location. Based on this distance map, the tendon is finally segmented and its shape refined by using active contours. The segmentation algorithm was tested on 90 images and experimental results demonstrate the accuracy of the proposed algorithm. The automatic segmentation was compared with an expert manual segmentation, and a mean error of 3.7 pixels and a standard deviation of 2 pixels were achieved, which are interested results for integration into future computer-assisted decision systems.
Download
Paper Nr: 15
Title:

A Framework for Creating Realistic Synthetic Fluorescence Microscopy Image Sequences
Authors:

Matsilele Mabaso, Daniel Withey and Bhekisipho Twala
Abstract: Fluorescence microscopy imaging is an important tool in modern biological research, allowing insights into the processes of biological systems. Automated image analysis algorithms help in extracting information from these images. Validation of the automated algorithms can be done with ground truth data based on manual annotations, or using synthetic data with known ground truth. Synthetic data avoids the need to annotate manually large datasets but may lack important characteristics of the real data. In this paper, we present a framework for the generation of realistic synthetic fluorescence microscopy image sequences of cells, based on the simulation of spots with realistic motion models, noise models, and with the use of real background from microscopy images. Our framework aims to close the gap between real and synthetic image sequences. To study the effect of real backgrounds, we compared three spot detection methods using our synthetic image sequences. The results show that the real background influences spot detection, reducing the effectiveness of the spot detection algorithms, indicating the value of synthetic images with a realistic background in system validation.
Download
Paper Nr: 18
Title:

Fully Automated Image Preprocessing for Feature Extraction from Knife-edge Scanning Microscopy Image Stacks - Towards a Fully Automated Image Processing Pipeline for Light Microscopic Images
Authors:

Shruthi Raghavan and Jaerock Kwon
Abstract: Knife-Edge Scanning Microscopy (KESM) stands out as a fast physical sectioning approach for imaging tissues at sub-micrometer resolution. To implement high-throughput and high-resolution, KESM images a tissue ribbon on the knife edge as the sample is being sectioned. This simultaneous sectioning and imaging approach has following benefits: (1) No image registration is required. (2) No manual job is required for tissue sectioning, placement or microscope imaging. However spurious pixels are present at the left and right side of the image, since the field of view of the objective is larger than the tissue width. The tissue region needs to be extracted from these images. Moreover, unwanted artifacts are introduced by KESM’s imaging mechanism, namely: (1) Vertical stripes caused by unevenly worn knife edge. (2) Horizontal artifacts due to vibration of the knife while cutting plastic embedded tissue. (3) Uneven intensity within an image due to knife misalignment. (4) Uneven intensity levels across images due to the variation of cutting speed. This paper outlines an image processing pipeline for extracting features from KESM images and proposes an algorithm to extract tissue region from physical sectioning-based light microscope images like KESM data for automating feature extraction from these data sets.
Download
Paper Nr: 23
Title:

EGFR-targeting Peptide Conjugated pH-sensitive Micelles as a Potential Drug Carrier for Photodynamic Detection and Therapy of Cancer
Authors:

Cheng-Liang Peng, Yuan-I Chen, Ying-Hsia Shih, Tsai-Yueh Luo and Ming-Jium Shieh
Abstract: Multifunctional theranostics have recently been intensively explored to optimize the efficacy and safety. Herein, we report multifunctional micelle that constructed from graft copolymer PEGMA-co-PDPA and diblock copolymer mPEG-b-PCL as the carrier of hydrophobic photosensitizer, chlorin e6 (Ce6) for simultaneous fluorescence imaging and photodynamic therapy. The functional inner core of PEGMA-co-PDPA exhibited pH stimulate to accelerate drug release under slightly acidic microenvironments of tumors and the outer shell of micelles with epidermal growth factor receptor (EGFR)-targeting GE11 peptides for active targeting of EGFR-overexpressing cancer cells. The results demonstrate that GE11-conjugated chlorin e6-loaded micelles (GE11-Ce6-micelles) with particle size around 100 nm and the micelles had well defined core shell structure which was evaluated by TEM. In the in vitro cellular uptake studies, GE11-Ce6-micelles exhibited a higher amount of intracellular uptake of chlorin e6 in HCT116 cancer cells (EGFR high expression) via receptor-mediated endocytosis, in contrast with the time-dependent passive uptake of the non-targeted Ce6-micelles, thereby providing a effective photocytotoxic effect on the HCT116 cancer cells. In vivo study revealed that GE11-Ce6-micelles exhibited tumor targeting for photodynamic detection and excellent inhibition on tumor growth after irradiation, indicating that GE11-Ce6-micelles could be successfully applied to the effective fluorescence imaging and photodynamic therapy of cancer.
Download
Paper Nr: 25
Title:

Supporting Novice Prehospital Transcranial Ultrasound Scanning for Brain Haemorrhage
Authors:

Leila Eadie, Luke Regan, Ashish MacAden and Philip Wilson
Abstract: Traumatic brain injury is a significant problem due to difficulties in early diagnosis in the field. Computed tomography is the gold standard for detecting brain haemorrhage, but scanners are bulky and expensive. A cheap, portable scanner such as transcranial ultrasound (TCUS) could allow early triage and intervention. Transmitting images to remote experts for diagnosis means TCUS could be used by any minimally trained person in the field. We propose a virtual 3-dimensional model of the head which shows which areas of the brain have been imaged already, where the probe currently is, and where still needs to be covered in order to generate a complete scan. Using sensors to measure the position and rotation of the TCUS transducer, we can link this to the 3D model of the head and visually display which areas have been imaged. The images can be analysed and composited to form a personalised 3D scan with maximal coverage of the brain, which can be transmitted for diagnostic review, reducing data loss compared with streaming ongoing images. Initial testing of the software has been performed in healthy volunteers and further testing is planned in patients with brain haemorrhage.
Download
Paper Nr: 29
Title:

Feasibility of Eye-tracking based Glasses-free 3D Autostereoscopic Display Systems for Medical 3D Images
Authors:

Dongwoo Kang, Seok Lee, Hyoseok Hwang, Juyong Park, Jingu Heo, Byongmin Kang, Jin-Ho Lee, Yoonsun Choi, Kyuhwan Choi and Dongkyung Nam
Abstract: Medical image diagnosis processes with stereoscopic depth by 3D display have not been developed widely yet and remain understudied Many stereoscopic displays require glasses that are inappropriate for use in clinical diagnosis/explanation/operating processes in hospitals. An eye-tracking based glasses-free three-dimensional autostereoscopic display monitor system has been developed, and its feasibility for medical 3D images was investigated, as a cardiac CT 3D navigator. Our autostereoscopic system uses slit-barrier with BLU, and it is combined with our vision-based eye tracking system to display 3D images. Dynamic light field rendering technique is applied with the 3D coordinates calculated by the eye-tracker, in order to provide a single viewer the best 3D images with less x-talk. To investigate the feasibility of our autostereoscopic system, 3D volume was rendered from 3D coronary CTA images (512 by 512 by 400). One expert reader identified the three main artery structures (LAD, LCX and RCA) in shorter time than existing 2D display. The reader did not report any eye fatigue or discomfort. In conclusion, we proposed a 3D cardiac CT navigator system with a new glasses-free 3D autostereoscopy, which may improve diagnosis accuracy and fasten diagnosis process.
Download
Paper Nr: 7
Title:

Automated Segmentation of Tumours in MRI Brain Scans
Authors:

Ali M. Hasan, Farid Meziane and Mohammad Abd Kadhim
Abstract: The research reported in this paper concerns the development of a novel automated algorithm to identify and segment brain tumours in MRI scans. The input is the patient's scan slices and the output is a subset of the slices that includes the tumour. The proposed method is called Bounding 3D Box Based Genetic Algorithm (BBBGA) and is based on the use of Genetic Algorithm (GA) to search for the most dissimilar regions between the left and right hemispheres of the brain. The process involves randomly generating a hundred of 3D boxes with different sizes and locations in the left hemisphere of the brain and compared with the corresponding 3D boxes in the right hemisphere of the brain through the objective function. These 3D boxes are moved and updated during the iterations of the GA towards the region of maximum dissimilarity between the two hemispheres which represent the approximate position of the tumour. The dataset includes 88 pathological patients provided by the MRI Unit of Al-Kadhimiya Teaching Hospital in Iraq. The achieved accuracy of the BBBGA and 3D segmentation of the tumour were 95% and 90% respectively.
Download
Paper Nr: 12
Title:

Interactive Chan-Vese Approach with Random Walk for Medical Images Segmentation
Authors:

Mohammadreza Hosseini, Arcot Sowmya and Tomasz Bednarz
Abstract: In this paper, we present a novel interactive variational approach to image segmentation within a Chan-Vese framework. We propose a parameterized energy function that can be modified based on user input, and also incorporate in it a probabilistic term that defines reachability of a pixel from a user-selected `internal’ object pixel. The proposed approach shows promising improvement over automatic segmentation methods when applied to medical images.
Download
Paper Nr: 14
Title:

Unsupervised Clustering of Hyperspectral Images of Brain Tissues by Hierarchical Non-negative Matrix Factorization
Authors:

Bangalore Ravi Kiran, Bogdan Stanciulescu and Jesus Angulo
Abstract: Hyperspectral images of high spatial and spectral resolutions are employed to perform the challenging task of brain tissue characterization and subsequent segmentation for visualization of in-vivo images. Each pixel is high-dimensional spectrum. Working on the hypothesis of pure-pixels on account of high spectral resolution, we perform unsupervised clustering by hierarchical non-negative matrix factorization to identify the pure-pixel spectral signatures of blood, brain tissues, tumor and other materials. This subspace clustering was further used to train a random forest for subsequent classification of test set images constituent of in-vivo and ex-vivo images. Unsupervised hierarchical clustering helps visualize tissue structure in in-vivo test images and provides a inter-operative tool for surgeons. Furthermore the study also provide a preliminary study of the classification and sources of errors in the classification process.
Download
Paper Nr: 22
Title:

Diagnostics of Optic Nerve Head Pathologies using Structural Analysis of Eye Ultrasound B-scan Images
Authors:

A. Kriščiukaitis, V. Valuckis, A. Kybartaitė-Žilienė and L. Kriaučiūnienė
Abstract: Optic nerve head drusen are congenital and developmental anomalies in a form of calcific degeneration in some of axons of the optic nerve head. Diagnostic difficulties may be encountered when drusen are buried deep within the nerve tissue in the optic nerve head, as they can resemble optic disc swelling or other pathologies. Diagnosing optical disc drusen correctly is important to avoid unnecessary work-up and to avoid overlooking potential serious conditions such as true papilledema. We propose the method based on structural analysis of the eye B-scan images combined with mathematical morphology to reveal valuable estimates reflecting pathogenic changes in the optic nerve and surrounding structures for improvement of diagnostic quality.
Download
Paper Nr: 24
Title:

Processing Magnetic Resonance Image Features with One-class Support Vector Machines - Investigation of the Autism Spectrum Disorder Heterogeneity
Authors:

Ilaria Gori, Alessia Giuliano, Piernicola Oliva, Michela Tosetti, Filippo Muratori, Sara Calderoni and Alessandra Retico
Abstract: Support Vector Machine (SVM) classifiers are widely used to analyse features extracted from brain MRI data to identify useful biomarkers of pathology in several disease conditions. They are trained to distinguish patients from healthy control subjects by making a binary classification of image features extracted by image processing algorithms. This task is particularly challenging when dealing with psychiatric disorders, as the reported neuroanatomical alterations are often very small and quite un-replicated within different studies. Subtle signs of pathology are difficult to catch especially in extremely heterogeneous conditions such as Autism Spectrum Disorders (ASD). We propose the use of the One-Class Classification (OCC) or Data Description method that, in contrast with two-class classification, is based on a description of one class of objects only. Then, new examples are tested for their similarity to the examples of this target class, end eventually considered as outliers. The application of the OCC to features extracted from brain MRI of children affected by ASD and control subjects demonstrated that a common pattern of features characterize the ASD population.
Download
Paper Nr: 26
Title:

A Semiautomatic Image Processing Tool to Measure Small Structures in Magnetic Resonance Images of the Brain at 7 Tesla - Application to Hippocampus Subfields of Patients with Mild Cognitive Impairment
Authors:

Alessandra Retico, Graziella Donatelli, Mauro Costagli, Laura Biagi, Maria Evelina Fantacci, Daniela Frosini, Gloria Tognoni, Mirco Cosottini and Michela Tosetti
Abstract: The current availability of Magnetic Resonance (MR) systems that operate at ultra high magnetic field (≥ 7 Tesla) allows the representation of anatomical structures at sub-millimeter resolution. Interestingly, small structures of the brain, such as the subfields of the hippocampus, the inner structures of the basal ganglia and of the brainstem become visible. Suitable software packages that allow analyzing and measuring such small structures are not currently readily available. We developed a semi-automated procedure to measure the thickness of the stratum radiatum and lacunosum-moleculare (SRLM) of the hippocampus. The change in the thickness of this subfield of the hippocampal formation is supposed to have a role in the pathological cognitive decline. Once we developed and validated the semiautomatic procedure on the 7T high-resolution T2*-weighted images of a healthy volunteer, we carried out a preliminary study on a population affected by Mild Cognitive Impairment to investigate the correlations of the SRLM thickness with the clinical scores of the patients, e.g. the Mini-Mental State Examination score and the Free and Cued Selective Reminding Test.
Download
Paper Nr: 28
Title:

A Support Vector Machine based Prediction Model for Discrimination of Malignant Pulmonary Nodules from Benign Nodules
Authors:

Yan Wu, Emmanuel Zachariah, Judith Amorosa, Anjani Naidu, Mina L. Labib, Jamil Shaikh, Donna Eckstein, Sinae Kim, John E. Langenfeld, Joseph Aisner, John L. Nosher, Robert S. DiPaola and David J. Foran
Abstract: Lung cancer is the leading cause of cancer death in the United States and worldwide. Most patients are diagnosed at an advanced stage, usually stage III or IV. Identification of lung cancer patients at an early stage might enable oncologists to surgically remove the tumors. Currently, low dose CT scans are used to identify the malignant nodules in high risk patients. However, screening CT scans yield a high rate of false-positive results. A prediction model was developed for improved discrimination of malignant nodules from benign nodules in patients who underwent lung screening CT. CT images and clinical outcomes of 39 patients were obtained from the National Lung Screening Trial (NLST), National Cancer Institute, National Institute of Health. Images were analyzed to extract computational features relevant to malignancy prediction. A Support Vector Machine (SVM) based model was developed to predict the malignancy of nodules. During pilot studies, our model achieved the following prediction performance: accuracy of 0.74, sensitivity of 0.85, and specificity of 0.61.
Download